[PATCH] workqueue: combine judgments in the loop of maybe_create_worker

[PATCH] workqueue: combine judgments in the loop of maybe_create_worker

[liusong]

From: Liu Song <liu.song11@zte.com.cn>

Although the 'need_to_create_worker' has been determined to be
true before entering the function. However, adjusting the order
of judgment can combine two judgments in the loop. Also improve
the matching between logical operations and function naming.

Signed-off-by: Liu Song <liu.song11@zte.com.cn>
---
 kernel/workqueue.c | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)

diff --git a/kernel/workqueue.c b/kernel/workqueue.c
index 78b192071ef7..3c7c90d70dcf 100644
--- a/kernel/workqueue.c
+++ b/kernel/workqueue.c
@@ -1969,13 +1969,10 @@ __acquires(&pool->lock)
 	mod_timer(&pool->mayday_timer, jiffies + MAYDAY_INITIAL_TIMEOUT);
 
 	while (true) {
-		if (create_worker(pool) || !need_to_create_worker(pool))
+		if (!need_to_create_worker(pool) || create_worker(pool))
 			break;
 
 		schedule_timeout_interruptible(CREATE_COOLDOWN);
-
-		if (!need_to_create_worker(pool))
-			break;
 	}
 
 	del_timer_sync(&pool->mayday_timer);
-- 
2.17.1

Re: [PATCH] workqueue: combine judgments in the loop of maybe_create_worker

[Tejun Heo]

On Wed, Sep 12, 2018 at 09:58:29PM +0800, liusong@tom.com wrote:
> From: Liu Song <liu.song11@zte.com.cn>
> 
> Although the 'need_to_create_worker' has been determined to be
> true before entering the function. However, adjusting the order
> of judgment can combine two judgments in the loop. Also improve
> the matching between logical operations and function naming.
> 
> Signed-off-by: Liu Song <liu.song11@zte.com.cn>

idk, this doesn't really improve anything.

Thanks.

-- 
tejun

Re: [PATCH] workqueue: combine judgments in the loop ofmaybe_create_worker

[Lai Jiangshan]

On Thu, Sep 13, 2018 at 9:51 AM <liu.song11@zte.com.cn> wrote:
>
> >> From: Liu Song <liu.song11@zte.com.cn>
> >>
> >> Although the 'need_to_create_worker' has been determined to be
> >> true before entering the function. However, adjusting the order
> >> of judgment can combine two judgments in the loop. Also improve
> >> the matching between logical operations and function naming.
> >>
> > >Signed-off-by: Liu Song <liu.song11@zte.com.cn>
> >
> >idk, this doesn't really improve anything.
> >
> >Thanks.
> >
> >--
> >tejun
>
> Hi, tejun
> Indeed, the logic of the code has not changed.

The logic of the code is changed in the patch when create_worker()
happened to be failed.

Failed create_worker() doesn't always result to COOLDOWN due to
the possible condition changes. if (!need_to_create_worker(pool)),
it dosn't need to COOLDOWN.



> The idea that I made this change is
> that when reading the workqueue code, from 'maybe_create_worker' naming, it is
> possible to create a worker, but the implementation will definitely create a worker,
> which makes me somewhat confused. After analysis, I found the determine whether
> to create a worker before the function call. So this patch seems to be able to reduce
> the confusion between the function naming and implementation.
>
> Thanks
> Best regards